Owly

Owly is the working prototype of Team Gefahrenherde at the HackTheHouse hackathon.

With this project we're showcasing how your environment, specifically your home, can become part of your everyday interactions. This means you're not living in your home anymore, but with it. Using a variety of technologies, we recognize people, situations that require action and even act accordingly in urgent cases.

In this hack we're demoing a system that uses a Microsoft Kinect to recognize unattended children in the kitchen and signal an alert by setting a Philips Hue on red, locking the connected oven and turning off a CUL compatible RC power socket to which a blender was connected. Currently this runs on a Linux Box, however we have managed to get the system to run on a Raspberry Pi 2 as well, albeit at higher latencies. We use the same infrastructure, augmented with a WunderBar gyroscope connected to the heat switch on an oven to detect wether it is turned on or not, to warn people when they leave the house. A working demo can be seen in the attached video.

We've also connected a WunderBar proximity sensor to a window, and integrating measurements from a Nest Thermostat, we warn the user when she has opened the bathroom window and has not turned down the heat just before or just after that, to save energy.

Furthermore, with the help of a microphone, we can detect when certain home appliances are done (the ones that are not connected to the internet), by listening for specific noises they make. In the demo we automatically send a tweet when the washing machine plays its done-melody.

The system will be designed to be expandable to many more input sources in order to give people an immersive experience of a truly connected home, instead of being forced to lock-in with a specific vendor. We envision a system that'll be able to tell people's behaviors and habits, and support them in their everyday living at home -- just like a family member would.

Detailed instructions to setup the whole thing, including hardware setup and source code, will be published soon.

The pitchdeck can be accessed here, where we also accredit all graphics that we used:

http://de.slideshare.net/Gefahrenherde/gefahrenherde

The video of the presentation can be seen here: